Casting Simulation for Local Mechanical Property Prediction

The patent creates virtual copies of the casting process through numerical simulation, generating multiple virtual casting scenarios that replicate real process variability. These virtual copies enable statistical analysis without requiring multiple physical experiments, thus improving prediction accuracy while avoiding the complexity of extensive physical testing.

The patent replaces physical experimentation and mechanical measurement systems with numerical simulation and statistical computing. By substituting the mechanical/physical prediction approach with a computational statistical model, the system achieves higher reliability through probability distributions while managing complexity through algorithmic processing.

The patent applies local quality by providing spatially resolved probability distributions of mechanical properties at different locations within the casting. This enables designers to apply safety factors only where statistically necessary based on local property variations and stress concentrations, rather than uniformly across the entire part, thus maintaining reliability while reducing unnecessary weight.

The patent changes the parameter representation from deterministic single values to probability distributions. This transformation allows for more nuanced reliability assessment, enabling optimization of safety margins based on statistical confidence levels rather than conservative uniform factors, thereby reducing weight while maintaining required reliability.

The patent performs preliminary virtual validation through numerical simulation before physical production. By conducting process simulation and statistical analysis in the virtual domain beforehand, the system provides confidence in casting properties without requiring extensive physical sampling, thus reducing pre-production time while maintaining reliability assessment capability.

The patent uses virtual copies of the casting process and material behavior to replace physical sampling. These numerical models replicate the casting process and predict mechanical properties with statistical confidence, eliminating the need for time-consuming physical prototypes and extensive sampling while maintaining validation confidence.

The patent enables local optimization by providing spatially varying probability distributions of mechanical properties. This allows lightweight design where material thickness and safety margins are optimized locally based on actual predicted property distributions and local stress states, rather than applying uniform conservative margins throughout, thus reducing weight while maintaining failure resistance.

The patent incorporates feedback loops where simulation results feed into design optimization. The statistical prediction of mechanical properties provides feedback on the actual reliability of lightweight designs, enabling iterative optimization that reduces weight while ensuring failure risk remains below acceptable thresholds through statistically informed safety margins.